Assessment And Prediction Of Soil Quality In Small Watershed Areas Of The Typical Black Soil Region In Northeast China

The environmental effects of soil erosion in the black soil area of northeast China are mainly reflected in degradation of soil quality,thinning of black soil thickness and reduction of grain production capacity,which seriously threaten national food security.Carrying out soil quality assessment research in black soil areas and exploring the establishment of regional soil quality prediction models can provide important scientific basis for protecting black soil resources and comprehensively controlling soil erosion.The research area we chosen is the typical small watershed of Heshan farm of Jiusan agricultural reclamation administration of Heilongjiang province.We thoroughly analyzed the characteristics and spatial distribution trends of main soil indexes in 3 layers of 61 sample sites.Then we used total data set(TDS)and minimum data set(MDS)combined with Nemoro quality index(NQI)and integrated quality index(IQI)to evaluate the quality of three layers of soil by four methods.The applicability of the Nemoro quality index and the integrated quality index were compared by efficiency ratio and sensitivity index.Moreover,we constructed a soil quality prediction model based on terrain parameters using random forests and verify its accuracy.The results can provide a basis for the selection of minimum dataset and soil quality algorithm model,and the study preliminarily establish a model for predicting soil quality by random forest based on terrain parameters.The main results are as follows:(i)The average soil bulk density of 0～10 cm,10～20 cm and 20～30 cm in the watershed increased with the increase of soil depth,and the average field capacity and total porosity in the field decreased with the increase of soil depth.There were highly significant negative correlations between bulk density and field capacity,bulk density and total porosity,and highly significant negative correlations between soil sand content and silt content,and soil sand content and clay content.The average soil p H is 5.29,which indicates that the soil is weakly acidic and decreases with the deepening of the soil layer.The mean values of soil organic matter,total nitrogen,available phosphorus and available potassium were 43.83 g/kg,2.10 g/kg,28.30 mg/kg and 106.34 mg/kg,respectively,and the contents decreased with soil depth.The cation exchange capacity was negatively correlated with the sand content,The activities of soil acid phosphatase and urease were positively correlated with the contents of total nitrogen,available phosphorus and available potassium.There was also a correlation between soil acid phosphatase and urease activity.The content of clay and silt particles tended to increase gradually along the decreasing contour,and the content of total nitrogen,available phosphorus,available potassium,cation exchange capacity,soil peroxidase activity and urease activity were all higher at the watershed outlet.The overall distribution pattern of soil quality index was higher at the watershed outlet and near the second forest belt in the study area,and lower in the southwestern part of the watershed.(ii)Based on the combined ranking of ER and SI,the IQI algorithm model can indicate the change of soil quality index better than NQI,and IQI is more suitable for soil quality evaluation in the study area.Based on principal component analysis MDS of 0-10cm soil layer was determined as silt content,field water capacity,acid phosphatase activity,total nitrogen,p H,available phosphorus and available potassium;MDS of 10-20 cm and20-30 cm soil layers were determined as field water capacity,silt content,total nitrogen,and p H.The F-tests for both the MDS-IQI and TDS-IQI fitted equations reached the P<0.001 highly significant level,indicating that the choice of MDS was reasonable.The integrated quality index calculation showed that the soil quality index ranged between 0.24and 0.67 for 0-10 cm,between 0.22 and 0.66 for 10-20 cm,and between 0.14 and 0.62 for20-30 cm,with low and lower-levels of soil quality dominating the sampling sites in the watershed.The percentage of medium level and high level quality soils decreased gradually as the soil depth increased.Soil quality index is strongly influenced by erosion deposition in the watershed,and soil erosion has a significant negative effect on SQI.(iii)We used a random forest model to predict soil quality indices based on topographic parameters and evaluated the model performance using mean absolute error(MAE),root mean square error(RMSE)and coefficient of determination(R~2),and the results showed that the predicted indices were generally consistent with those based on the integrated quality index algorithm.The relative importance results derived from the random forest showed that distance from the forest zone was identified as the most important driver of spatial variability in soil quality,accounting for 57.37%,slope followed by 11.12%,and elevation,profile curvature,topographic moisture index,and topographic relief had similar importance,all ranging from 6%to 7%.